Install snappy library

add create-load-data.sh

.gitignore

@@ -1,10 +1,12 @@
 *~
+.*.swp
 derby.log
 thirdparty
 cscope.files
 cscope.out
 org.eclipse.jdt.core.prefs
 hive_benchmark_results.txt
+testdata/data/test-warehouse
 
 pprof.out
 

CMakeLists.txt

@@ -92,6 +92,16 @@ find_package(Llvm REQUIRED)
 include_directories(${LLVM_INCLUDE_DIR})
 set(LIBS ${LIBS} ${LLVM_MODULE_LIBS})
 
+# find Snappy headers and libs
+find_package(Snappy REQUIRED)
+include_directories(${SNAPPY_INCLUDE_DIR})
+set(LIBS ${LIBS} ${SNAPPY_LIBRARIES})
+add_library(snappy STATIC IMPORTED)
+set_target_properties(snappy PROPERTIES IMPORTED_LOCATION "${SNAPPY_LIBRARY}")
+
+message(STATUS ${SNAPPY_INCLUDE_DIR})
+message(STATUS ${SNAPPY_LIBRARY})
+
 # compile these subdirs using their own CMakeLists.txt
 add_subdirectory(common/function-registry)
 add_subdirectory(common/thrift)

be/src/exec/CMakeLists.txt

@@ -37,5 +37,6 @@ target_link_libraries(Exec
 target_link_libraries(Exec
   ${JAVA_JVM_LIBRARY}
   ${HDFS_LIBS}
-  -lz -lbz2 -lsnappy
+  snappy
+  -lz -lbz2
 )

be/src/exec/hdfs-sequence-scanner.cc

@@ -14,7 +14,7 @@
 // Compression libraries
 #include <zlib.h>
 #include <bzlib.h>
-#include <snappy.h>
+#include "snappy.h"
 
 using namespace std;
 using namespace boost;
@@ -226,9 +226,9 @@ Status HdfsSequenceScanner::InitCurrentScanRange(RuntimeState* state,
   end_of_scan_range_ = scan_range->length + scan_range->offset;
   unbuffered_byte_stream_ = byte_stream;
 
-  // If the file is blocked compressed then we don't want to double buffer
+  // If the file is blocked-compressed then we don't want to double buffer
   // the compressed blocks.  In that case we read meta information in
-  // filesystem block sizes (4kb) otherwise we read large chunks (1Mb)
+  // filesystem block sizes (4KB) otherwise we read large chunks (1MB)
   // and pick meta data and data from that buffer.
   buffered_byte_stream_.reset(new BufferedByteStream(
       unbuffered_byte_stream_,

be/src/exprs/CMakeLists.txt

@@ -62,9 +62,10 @@ target_link_libraries(expr-test
   Opcode
   Exprs
   gtest
+  snappy
   ${Boost_LIBRARIES}
   ${LLVM_MODULE_LIBS}
-  -lz -lbz2 -lsnappy
+  -lz -lbz2 
 )
 
 add_test(expr-test ${BUILD_OUTPUT_ROOT_DIRECTORY}/exprs/expr-test)

be/src/runtime/CMakeLists.txt

@@ -32,7 +32,7 @@ target_link_libraries(Runtime
   Exec
   TestUtil
   ${Boost_LIBRARIES}
-  -lz -lbz2 -lsnappy
+  -lz -lbz2
 )
 
 add_executable(mem-pool-test

be/src/service/CMakeLists.txt

@@ -46,7 +46,7 @@ target_link_libraries(backend
   gflagsstatic
 #  tcmallocstatic
   pprofstatic
-  -lz -lbz2 -lsnappy
+  -lz -lbz2
 )
 
 add_executable(runquery 
@@ -80,12 +80,13 @@ target_link_libraries(runquery
   Exprs
   Opcode
   Exprs
+  snappy
   ${Boost_LIBRARIES}  
   ${LLVM_MODULE_LIBS}
   gflagsstatic
   tcmallocstatic
   pprofstatic
-  -lz
+  -lz -lbz2
 )
 
 add_executable(impalad
@@ -112,6 +113,7 @@ target_link_libraries(impalad
   Opcode
   Exprs
   CodeGen
+  snappy
   ${Boost_LIBRARIES}  
   ${LLVM_MODULE_LIBS}
   gflagsstatic

be/src/testutil/CMakeLists.txt

@@ -19,7 +19,7 @@ target_link_libraries(TestUtil
   ImpalaThrift
   glogstatic
   gflagsstatic  
-  -lz -lbz2 -lsnappy
+  -lz -lbz2
 )
 
 add_executable(query-jitter

bin/impala-config.sh

@@ -55,6 +55,8 @@ export LIBHDFS_OPTS="-Djava.library.path=${HADOOP_HOME}/lib/native/"
 
 export ARTISTIC_STYLE_OPTIONS=$IMPALA_BE_DIR/.astylerc
 
+export JAVA_LIBRARY_PATH=${IMPALA_HOME}/thirdparty/snappy-1.0.5/build/lib
+
 CLASSPATH=$IMPALA_FE_DIR/target/dependency:$CLASSPATH
 CLASSPATH=$IMPALA_FE_DIR/target/classes:$CLASSPATH
 CLASSPATH=$IMPALA_FE_DIR/src/test/resources:$CLASSPATH

buildall.sh

@@ -139,6 +139,14 @@ then
 fi
 make -j4
 
+# Build Snappy
+cd $IMPALA_HOME/thirdparty/snappy-1.0.5
+if [ $config_action -eq 1 ]
+then
+  ./configure --with-pic --prefix=$IMPALA_HOME/thirdparty/snappy-1.0.5/build
+fi
+make install
+
 # cleanup FE process
 $IMPALA_HOME/bin/clean-fe-processes.py
 

cmake_modules/FindSnappy.cmake

@@ -0,0 +1,53 @@
+# - Find SNAPPY (snappy.h, libsnappy.a, libsnappy.so, and libsnappy.so.1)
+# This module defines
+#  SNAPPY_INCLUDE_DIR, directory containing headers
+#  SNAPPY_LIBS, directory containing gflag libraries
+#  SNAPPY_STATIC_LIB, path to libgflags.a
+#  SNAPPY_FOUND, whether gflags has been found
+
+set(SNAPPY_SEARCH_HEADER_PATHS  
+  ${CMAKE_SOURCE_DIR}/thirdparty/snappy-1.0.5/build/include
+)
+
+set(SNAPPY_SEARCH_LIB_PATH
+  ${CMAKE_SOURCE_DIR}/thirdparty/snappy-1.0.5/build/lib
+)
+
+set(SNAPPY_INCLUDE_DIR 
+  ${CMAKE_SOURCE_DIR}/thirdparty/snappy-1.0.5/build/include
+)
+
+find_library(GTEST_LIBRARY NAMES gtest
+  PATHS ${CMAKE_SOURCE_DIR}/thirdparty/gtest-1.6.0
+        NO_DEFAULT_PATH
+  DOC   "Google's framework for writing C++ tests (gtest)"
+)
+find_library(SNAPPY_LIBRARY NAMES snappy
+  PATHS ${SNAPPY_SEARCH_LIB_PATH}
+        NO_DEFAULT_PATH
+  DOC   "Google's snappy compression library"
+)
+
+if (SNAPPY_LIB_PATH)
+  set(SNAPPY_FOUND TRUE)
+  set(SNAPPY_LIBS ${SNAPPY_SEARCH_LIB_PATH})
+  set(SNAPPY_STATIC_LIB ${SNAPPY_SEARCH_LIB_PATH}/libgflags.a)
+else ()
+  set(SNAPPY_FOUND FALSE)
+endif ()
+
+if (SNAPPY_FOUND)
+  if (NOT SNAPPY_FIND_QUIETLY)
+    message(STATUS "Snappy Found in ${SNAPPY_SEARCH_LIB_PATH}")
+  endif ()
+else ()
+  message(STATUS "Snappy includes and libraries NOT found. "
+    "Looked for headers in ${SNAPPY_SEARCH_HEADER_PATH}, "
+    "and for libs in ${SNAPPY_SEARCH_LIB_PATH}")
+endif ()
+
+mark_as_advanced(
+  SNAPPY_INCLUDE_DIR
+  SNAPPY_LIBS
+  SNAPPY_STATIC_LIB
+)

fe/pom.xml

@@ -540,19 +540,7 @@
                     <exec executable="${env.IMPALA_HOME}/testdata/bin/create-hbase.sh">
                     </exec>
                     <!-- create Hive table and load Hive data -->
-                    <exec executable="${env.HIVE_HOME}/bin/hive">
+                    <exec executable="${env.IMPALA_HOME}/testdata/bin/create-load-data.sh">
-                      <arg value="-hiveconf"/>
-                      <arg value="hive.root.logger=WARN,console"/>
-                      <arg value="-v"/>
-                      <arg value="-f"/>
-                      <arg value="${project.basedir}/../testdata/bin/create.sql"/>
-                    </exec>
-                    <exec executable="${env.HIVE_HOME}/bin/hive">
-                      <arg value="-hiveconf"/>
-                      <arg value="hive.root.logger=WARN,console"/>
-                      <arg value="-v"/>
-                      <arg value="-f"/>
-                      <arg value="${project.basedir}/../testdata/bin/load.sql"/>
                     </exec>                    
                   </target>
                     <!-- split HBase tables -->

testdata/bin/create-benchmark.sql

@@ -46,3 +46,35 @@ CREATE TABLE Grep1GB_seq_snap (
   field string)
 partitioned by (chunk int)
 STORED AS SEQUENCEFILE;
+
+DROP TABLE IF EXISTS UserVisits_rc;
+CREATE TABLE UserVisits_rc (
+  sourceIP string,
+  destURL string,
+  visitDate string,
+  adRevenue float,
+  userAgent string,
+  cCode string,
+  lCode string,
+  sKeyword string,
+  avgTimeOnSite int)
+STORED AS RCFILE;
+
+DROP TABLE IF EXISTS UserVisits_rc_bzip;
+CREATE TABLE UserVisits_rc_bzip (
+  sourceIP string,
+  destURL string,
+  visitDate string,
+  adRevenue float,
+  userAgent string,
+  cCode string,
+  lCode string,
+  sKeyword string,
+  avgTimeOnSite int)
+STORED AS RCFILE;
+
+DROP TABLE IF EXISTS Grep1GB_rc;
+CREATE TABLE Grep1GB_rc (
+  field string)
+partitioned by (chunk int)
+STORED AS RCFILE;

testdata/bin/create-load-data.sh

@@ -0,0 +1,36 @@
+#!/bin/bash
+# Copyright (c) 2012 Cloudera, Inc. All rights reserved.
+
+if [ x${JAVA_HOME} == x ]; then
+  echo JAVA_HOME not set
+  exit -1
+fi
+
+${HIVE_HOME}/bin/hive -hiveconf hive.root.logger=WARN,console -v \
+  -f ${IMPALA_HOME}/testdata/bin/create.sql 
+if [ $? != 0 ]; then
+  echo CREATE FAILED
+  exit -1
+fi
+if [ -d ${IMPALA_HOME}/testdata/data/test-warehouse ] ; then 
+  # The data has already been created, just load it.
+  ${HIVE_HOME}/bin/hive -hiveconf hive.root.logger=WARN,console -v \
+    -f ${IMPALA_HOME}/testdata/bin/load.sql 
+  if [ $? != 0 ]; then
+    echo LOAD FAILED
+    exit -1
+  fi
+else
+  ${HIVE_HOME}/bin/hive -hiveconf hive.root.logger=WARN,console -v \
+    -f ${IMPALA_HOME}/testdata/bin/load-raw-data.sql
+  if [ $? != 0 ]; then 
+    echo RAW DATA LOAD FAILED
+    exit -1
+  fi
+  cd ${IMPALA_HOME}/testdata/data
+  hadoop fs -get /test-warehouse
+  if [ $? != 0 ]; then 
+    echo HADOOP GET FAILED
+    exit -1
+  fi
+fi

testdata/bin/create.sql

@@ -92,6 +92,18 @@ CREATE TABLE AllTypesSmall LIKE AllTypes;
 DROP TABLE IF EXISTS AllTypesSmall_rc;
 CREATE TABLE AllTypesSmall_rc LIKE AllTypes_rc;
 
+DROP TABLE IF EXISTS AllTypesSmall_rc_def;
+CREATE TABLE AllTypesSmall_rc_def LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AllTypesSmall_rc_gzip;
+CREATE TABLE AllTypesSmall_rc_gzip LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AllTypesSmall_rc_bzip;
+CREATE TABLE AllTypesSmall_rc_bzip LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AllTypesSmall_rc_snap;
+CREATE TABLE AllTypesSmall_rc_snap LIKE AllTypes_rc;
+
 DROP TABLE IF EXISTS AllTypesSmall_seq;
 CREATE TABLE AllTypesSmall_seq LIKE AllTypes_seq;
 
@@ -125,6 +137,18 @@ CREATE TABLE AllTypesError LIKE AllTypes;
 DROP TABLE IF EXISTS AlltypesError_rc;
 CREATE TABLE AllTypesError_rc LIKE AllTypes_rc;
 
+DROP TABLE IF EXISTS AlltypesError_rc_def;
+CREATE TABLE AllTypesError_rc_def LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AlltypesError_rc_gzip;
+CREATE TABLE AllTypesError_rc_gzip LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AlltypesError_rc_bzip;
+CREATE TABLE AllTypesError_rc_bzip LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AlltypesError_rc_snap;
+CREATE TABLE AllTypesError_rc_snap LIKE AllTypes_rc;
+
 DROP TABLE IF EXISTS AlltypesError_seq;
 CREATE TABLE AllTypesError_seq LIKE AllTypes_seq;
 
@@ -158,6 +182,18 @@ CREATE TABLE AllTypesErrorNoNulls LIKE AllTypes;
 DROP TABLE IF EXISTS AlltypesErrorNoNulls_rc;
 CREATE TABLE AllTypesErrorNoNulls_rc LIKE AllTypes_rc;
 
+DROP TABLE IF EXISTS AlltypesErrorNoNulls_rc_def;
+CREATE TABLE AllTypesErrorNoNulls_rc_def LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AlltypesErrorNoNulls_rc_gzip;
+CREATE TABLE AllTypesErrorNoNulls_rc_gzip LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AlltypesErrorNoNulls_rc_bzip;
+CREATE TABLE AllTypesErrorNoNulls_rc_bzip LIKE AllTypes_rc;
+
+DROP TABLE IF EXISTS AlltypesErrorNoNulls_rc_snap;
+CREATE TABLE AllTypesErrorNoNulls_rc_snap LIKE AllTypes_rc;
+
 DROP TABLE IF EXISTS AlltypesErrorNoNulls_seq;
 CREATE TABLE AllTypesErrorNoNulls_seq LIKE AllTypes_seq;
 
@@ -217,6 +253,18 @@ CREATE TABLE AllTypesAgg_rc (
 partitioned by (year int, month int, day int)
 STORED AS RCFILE;
 
+DROP TABLE IF EXISTS AllTypesAgg_rc_def;
+CREATE TABLE AllTypesAgg_rc_def LIKE AllTypesAgg_rc;
+
+DROP TABLE IF EXISTS AllTypesAgg_rc_gzip;
+CREATE TABLE AllTypesAgg_rc_gzip LIKE AllTypesAgg_rc;
+
+DROP TABLE IF EXISTS AllTypesAgg_rc_bzip;
+CREATE TABLE AllTypesAgg_rc_bzip LIKE AllTypesAgg_rc;
+
+DROP TABLE IF EXISTS AllTypesAgg_rc_snap;
+CREATE TABLE AllTypesAgg_rc_snap LIKE AllTypesAgg_rc;
+
 DROP TABLE IF EXISTS AllTypesAgg_seq;
 CREATE TABLE AllTypesAgg_seq (
   id int,
@@ -263,6 +311,18 @@ CREATE TABLE AllTypesAggNoNulls LIKE AllTypesAgg;
 DROP TABLE IF EXISTS AllTypesAggNoNulls_rc;
 CREATE TABLE AllTypesAggNoNulls_rc LIKE AllTypesAgg_rc;
 
+DROP TABLE IF EXISTS AllTypesAggNoNulls_rc_def;
+CREATE TABLE AllTypesAggNoNulls_rc_def LIKE AllTypesAgg_rc;
+
+DROP TABLE IF EXISTS AllTypesAggNoNulls_rc_gzip;
+CREATE TABLE AllTypesAggNoNulls_rc_gzip LIKE AllTypesAgg_rc;
+
+DROP TABLE IF EXISTS AllTypesAggNoNulls_rc_bzip;
+CREATE TABLE AllTypesAggNoNulls_rc_bzip LIKE AllTypesAgg_rc;
+
+DROP TABLE IF EXISTS AllTypesAggNoNulls_rc_snap;
+CREATE TABLE AllTypesAggNoNulls_rc_snap LIKE AllTypesAgg_rc;
+
 DROP TABLE IF EXISTS AllTypesAggNoNulls_seq;
 CREATE TABLE AllTypesAggNoNulls_seq LIKE AllTypesAgg_seq;
 
@@ -315,6 +375,18 @@ CREATE TABLE TestTbl_rc (
   zip int)
 STORED AS RCFILE;
 
+DROP TABLE IF EXISTS TestTbl_rc_def;
+CREATE TABLE TestTbl_rc_def LIKE TestTbl_rc;
+
+DROP TABLE IF EXISTS TestTbl_rc_gzip;
+CREATE TABLE TestTbl_rc_gzip LIKE TestTbl_rc;
+
+DROP TABLE IF EXISTS TestTbl_rc_bzip;
+CREATE TABLE TestTbl_rc_bzip LIKE TestTbl_rc;
+
+DROP TABLE IF EXISTS TestTbl_rc_snap;
+CREATE TABLE TestTbl_rc_snap LIKE TestTbl_rc;
+
 DROP TABLE IF EXISTS TestTbl_seq;
 CREATE TABLE TestTbl_seq (
   id bigint,
@@ -588,6 +660,18 @@ CREATE TABLE AllTypesAggMultiFiles_rc (
 partitioned by (year int, month int, day int)
 STORED AS RCFILE;
 
+DROP TABLE IF EXISTS AllTypesAggMultiFiles_rc_def;
+CREATE TABLE AllTypesAggMultiFiles_rc_def LIKE AllTypesAggMultiFiles_rc;
+
+DROP TABLE IF EXISTS AllTypesAggMultiFiles_rc_gzip;
+CREATE TABLE AllTypesAggMultiFiles_rc_gzip LIKE AllTypesAggMultiFiles_rc;
+
+DROP TABLE IF EXISTS AllTypesAggMultiFiles_rc_bzip;
+CREATE TABLE AllTypesAggMultiFiles_rc_bzip LIKE AllTypesAggMultiFiles_rc;
+
+DROP TABLE IF EXISTS AllTypesAggMultiFiles_rc_snap;
+CREATE TABLE AllTypesAggMultiFiles_rc_snap LIKE AllTypesAggMultiFiles_rc;
+
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq;
 CREATE TABLE AllTypesAggMultiFiles_seq (
   id int,
@@ -605,132 +689,28 @@ partitioned by (year int, month int, day int)
 STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_def;
-CREATE TABLE AllTypesAggMultiFiles_seq_def (
+CREATE TABLE AllTypesAggMultiFiles_seq_def LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_gzip;
-CREATE TABLE AllTypesAggMultiFiles_seq_gzip (
+CREATE TABLE AllTypesAggMultiFiles_seq_gzip LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_bzip;
-CREATE TABLE AllTypesAggMultiFiles_seq_bzip (
+CREATE TABLE AllTypesAggMultiFiles_seq_bzip LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_snap;
-CREATE TABLE AllTypesAggMultiFiles_seq_snap (
+CREATE TABLE AllTypesAggMultiFiles_seq_snap LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_record_def;
-CREATE TABLE AllTypesAggMultiFiles_seq_record_def (
+CREATE TABLE AllTypesAggMultiFiles_seq_record_def LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_record_gzip;
-CREATE TABLE AllTypesAggMultiFiles_seq_record_gzip (
+CREATE TABLE AllTypesAggMultiFiles_seq_record_gzip LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_record_bzip;
-CREATE TABLE AllTypesAggMultiFiles_seq_record_bzip (
+CREATE TABLE AllTypesAggMultiFiles_seq_record_bzip LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFiles_seq_record_snap;
-CREATE TABLE AllTypesAggMultiFiles_seq_record_snap (
+CREATE TABLE AllTypesAggMultiFiles_seq_record_snap LIKE AllTypesAggMultiFiles_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-partitioned by (year int, month int, day int)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart;
 CREATE TABLE AllTypesAggMultiFilesNoPart (
@@ -776,6 +756,18 @@ CREATE TABLE AllTypesAggMultiFilesNoPart_rc (
   timestamp_col timestamp)
 STORED AS RCFILE;
 
+DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_rc_def;
+CREATE TABLE AllTypesAggMultiFilesNoPart_rc_def LIKE AllTypesAggMultiFilesNoPart_rc;
+
+DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_rc_gzip;
+CREATE TABLE AllTypesAggMultiFilesNoPart_rc_gzip LIKE AllTypesAggMultiFilesNoPart_rc;
+
+DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_rc_bzip;
+CREATE TABLE AllTypesAggMultiFilesNoPart_rc_bzip LIKE AllTypesAggMultiFilesNoPart_rc;
+
+DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_rc_snap;
+CREATE TABLE AllTypesAggMultiFilesNoPart_rc_snap LIKE AllTypesAggMultiFilesNoPart_rc;
+
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq;
 CREATE TABLE AllTypesAggMultiFilesNoPart_seq (
   id int,
@@ -792,121 +784,29 @@ CREATE TABLE AllTypesAggMultiFilesNoPart_seq (
 STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_def;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_def (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_def LIKE AllTypesAggMultiFilesNoPart_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_gzip;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_gzip (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_gzip LIKE AllTypesAggMultiFilesNoPart_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_bzip;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_bzip (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_bzip LIKE AllTypesAggMultiFilesNoPart_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_snap;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_snap (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_snap LIKE AllTypesAggMultiFilesNoPart_seq;
-  id int,
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_record_def;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_def (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_def
-  id int,
+LIKE AllTypesAggMultiFilesNoPart_seq;
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_record_gzip;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_gzip (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_gzip
-  id int,
+LIKE AllTypesAggMultiFilesNoPart_seq;
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_record_bzip;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_bzip (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_bzip
-  id int,
+LIKE AllTypesAggMultiFilesNoPart_seq;
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;
 
 DROP TABLE IF EXISTS AllTypesAggMultiFilesNoPart_seq_record_snap;
-CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_snap (
+CREATE TABLE AllTypesAggMultiFilesNoPart_seq_record_snap
-  id int,
+LIKE AllTypesAggMultiFilesNoPart_seq;
-  bool_col boolean,
-  tinyint_col tinyint,
-  smallint_col smallint,
-  int_col int,
-  bigint_col bigint,
-  float_col float,
-  double_col double,
-  date_string_col string,
-  string_col string,
-  timestamp_col timestamp)
-STORED AS SEQUENCEFILE;

testdata/bin/load-raw-data.sql

@@ -138,6 +138,25 @@ SET hive.exec.compress.output=true;
 set mapred.output.compression.type=BLOCK;
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
 
+DROP TABLE IF EXISTS alltypeserror_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserror_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserror_rc_def';
+INSERT OVERWRITE TABLE alltypeserror_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserror_tmp;
+
 DROP TABLE IF EXISTS alltypeserror_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserror_seq_tmp (
   id STRING,
@@ -181,6 +200,25 @@ INSERT OVERWRITE TABLE alltypeserror_seq_tmp PARTITION (year, month)
 SELECT * FROM alltypeserror_tmp;
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
+DROP TABLE IF EXISTS alltypeserror_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserror_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserror_rc_gzip';
+INSERT OVERWRITE TABLE alltypeserror_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserror_tmp;
+
 DROP TABLE IF EXISTS alltypeserror_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserror_seq_tmp (
   id STRING,
@@ -224,6 +262,25 @@ INSERT OVERWRITE TABLE alltypeserror_seq_tmp PARTITION (year, month)
 SELECT * FROM alltypeserror_tmp;
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
+DROP TABLE IF EXISTS alltypeserror_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserror_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserror_rc_bzip';
+INSERT OVERWRITE TABLE alltypeserror_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserror_tmp;
+
 DROP TABLE IF EXISTS alltypeserror_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserror_seq_tmp (
   id STRING,
@@ -267,6 +324,25 @@ INSERT OVERWRITE TABLE alltypeserror_seq_tmp PARTITION (year, month)
 SELECT * FROM alltypeserror_tmp;
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
+DROP TABLE IF EXISTS alltypeserror_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserror_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserror_rc_snap';
+INSERT OVERWRITE TABLE alltypeserror_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserror_tmp;
+
 DROP TABLE IF EXISTS alltypeserror_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserror_seq_tmp (
   id STRING,
@@ -320,6 +396,26 @@ PARTITION (year=2009, month=1)
 PARTITION (year=2009, month=2)
 PARTITION (year=2009, month=3);
 
+ALTER TABLE AllTypesError_rc_def ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
+ALTER TABLE AllTypesError_rc_gzip ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
+ALTER TABLE AllTypesError_rc_bzip ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
+ALTER TABLE AllTypesError_rc_snap ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
 -- and for AllTypesError_Seq
 ALTER TABLE AllTypesError_seq ADD
 PARTITION (year=2009, month=1)
@@ -489,6 +585,33 @@ SET hive.exec.compress.output=true;
 set mapred.output.compression.type=BLOCK;
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
 
+DROP TABLE IF EXISTS alltypeserrornonulls_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserrornonulls_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserrornonulls_rc_def';
+
+set hive.exec.dynamic.partition=true;
+set hive.exec.dynamic.partition.mode=nonstrict;
+INSERT OVERWRITE TABLE alltypeserrornonulls_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserrornonulls_tmp;
+
+ALTER TABLE AllTypesErrorNoNulls_rc_def ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
 DROP TABLE AllTypesErrorNoNulls_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserrornonulls_seq_tmp (
   id STRING,
@@ -517,6 +640,33 @@ PARTITION (year=2009, month=2)
 PARTITION (year=2009, month=3);
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
+DROP TABLE IF EXISTS alltypeserrornonulls_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserrornonulls_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserrornonulls_rc_gzip';
+
+set hive.exec.dynamic.partition=true;
+set hive.exec.dynamic.partition.mode=nonstrict;
+INSERT OVERWRITE TABLE alltypeserrornonulls_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserrornonulls_tmp;
+
+ALTER TABLE AllTypesErrorNoNulls_rc_gzip ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
 DROP TABLE AllTypesErrorNoNulls_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserrornonulls_seq_tmp (
   id STRING,
@@ -545,6 +695,33 @@ PARTITION (year=2009, month=2)
 PARTITION (year=2009, month=3);
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
+DROP TABLE IF EXISTS alltypeserrornonulls_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserrornonulls_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserrornonulls_rc_bzip';
+
+set hive.exec.dynamic.partition=true;
+set hive.exec.dynamic.partition.mode=nonstrict;
+INSERT OVERWRITE TABLE alltypeserrornonulls_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserrornonulls_tmp;
+
+ALTER TABLE AllTypesErrorNoNulls_rc_bzip ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
 DROP TABLE AllTypesErrorNoNulls_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserrornonulls_seq_tmp (
   id STRING,
@@ -573,6 +750,33 @@ PARTITION (year=2009, month=2)
 PARTITION (year=2009, month=3);
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
+DROP TABLE IF EXISTS alltypeserrornonulls_rc_tmp;
+CREATE EXTERNAL TABLE alltypeserrornonulls_rc_tmp (
+  id STRING,
+  bool_col STRING,
+  tinyint_col STRING,
+  smallint_col STRING,
+  int_col STRING,
+  bigint_col STRING,
+  float_col STRING,
+  double_col STRING,
+  date_string_col STRING,
+  string_col STRING,
+  timestamp_col STRING)
+PARTITIONED BY (year INT, month INT)
+STORED AS RCFILE
+LOCATION '${hiveconf:hive.metastore.warehouse.dir}/alltypeserrornonulls_rc_snap';
+
+set hive.exec.dynamic.partition=true;
+set hive.exec.dynamic.partition.mode=nonstrict;
+INSERT OVERWRITE TABLE alltypeserrornonulls_rc_tmp PARTITION (year, month)
+SELECT * FROM alltypeserrornonulls_tmp;
+
+ALTER TABLE AllTypesErrorNoNulls_rc_snap ADD
+PARTITION (year=2009, month=1)
+PARTITION (year=2009, month=2)
+PARTITION (year=2009, month=3);
+
 DROP TABLE AllTypesErrorNoNulls_seq_tmp;
 CREATE EXTERNAL TABLE alltypeserrornonulls_seq_tmp (
   id STRING,
@@ -763,6 +967,22 @@ SET hive.exec.compress.output=true;
 set mapred.output.compression.type=BLOCK;
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
 
+INSERT OVERWRITE TABLE alltypes_rc_def partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypes;
+
+INSERT OVERWRITE TABLE alltypessmall_rc_def partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypessmall;
+
+INSERT OVERWRITE TABLE alltypesagg_rc_def partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesagg;
+
+INSERT OVERWRITE TABLE alltypesaggnonulls_rc_def partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesaggnonulls;
+
 INSERT OVERWRITE TABLE alltypes_seq_def partition (year, month)
 SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
 FROM alltypes;
@@ -780,6 +1000,22 @@ SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col,
 FROM alltypesaggnonulls;
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
+INSERT OVERWRITE TABLE alltypes_rc_gzip partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypes;
+
+INSERT OVERWRITE TABLE alltypessmall_rc_gzip partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypessmall;
+
+INSERT OVERWRITE TABLE alltypesagg_rc_gzip partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesagg;
+
+INSERT OVERWRITE TABLE alltypesaggnonulls_rc_gzip partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesaggnonulls;
+
 INSERT OVERWRITE TABLE alltypes_seq_gzip partition (year, month)
 SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
 FROM alltypes;
@@ -797,6 +1033,22 @@ SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col,
 FROM alltypesaggnonulls;
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
+INSERT OVERWRITE TABLE alltypes_rc_bzip partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypes;
+
+INSERT OVERWRITE TABLE alltypessmall_rc_bzip partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypessmall;
+
+INSERT OVERWRITE TABLE alltypesagg_rc_bzip partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesagg;
+
+INSERT OVERWRITE TABLE alltypesaggnonulls_rc_bzip partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesaggnonulls;
+
 INSERT OVERWRITE TABLE alltypes_seq_bzip partition (year, month)
 SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
 FROM alltypes;
@@ -814,6 +1066,22 @@ SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col,
 FROM alltypesaggnonulls;
 
 SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
+INSERT OVERWRITE TABLE alltypes_rc_snap partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypes;
+
+INSERT OVERWRITE TABLE alltypessmall_rc_snap partition (year, month)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
+FROM alltypessmall;
+
+INSERT OVERWRITE TABLE alltypesagg_rc_snap partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesagg;
+
+INSERT OVERWRITE TABLE alltypesaggnonulls_rc_snap partition (year, month, day)
+SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day
+FROM alltypesaggnonulls;
+
 INSERT OVERWRITE TABLE alltypes_seq_snap partition (year, month)
 SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month
 FROM alltypes;
@@ -944,49 +1212,93 @@ insert into table alltypesaggmultifilesnopart_seq   SELECT id, bool_col, tinyint
 insert into table alltypesaggmultifilesnopart_seq   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET hive.exec.compress.output=true; 
+set mapred.output.compression.type=BLOCK;
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
+
+insert into table alltypesaggmultifilesnopart_rc_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
+insert into table alltypesaggmultifilesnopart_rc_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
+insert into table alltypesaggmultifilesnopart_rc_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
+insert into table alltypesaggmultifilesnopart_rc_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
 insert into table alltypesaggmultifilesnopart_seq_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
+insert into table alltypesaggmultifilesnopart_rc_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
+insert into table alltypesaggmultifilesnopart_rc_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
+insert into table alltypesaggmultifilesnopart_rc_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
+insert into table alltypesaggmultifilesnopart_rc_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
 insert into table alltypesaggmultifilesnopart_seq_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
+insert into table alltypesaggmultifilesnopart_rc_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
+insert into table alltypesaggmultifilesnopart_rc_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
+insert into table alltypesaggmultifilesnopart_rc_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
+insert into table alltypesaggmultifilesnopart_rc_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
 insert into table alltypesaggmultifilesnopart_seq_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
+insert into table alltypesaggmultifilesnopart_rc_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
+insert into table alltypesaggmultifilesnopart_rc_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
+insert into table alltypesaggmultifilesnopart_rc_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
+insert into table alltypesaggmultifilesnopart_rc_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
 insert into table alltypesaggmultifilesnopart_seq_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
 
+set mapred.output.compression.type=RECORD;
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
 insert into table alltypesaggmultifilesnopart_seq_record_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_record_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_record_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_record_def   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
 insert into table alltypesaggmultifilesnopart_seq_record_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_record_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_record_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_record_gzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
 insert into table alltypesaggmultifilesnopart_seq_record_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_record_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_record_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_record_bzip   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
 insert into table alltypesaggmultifilesnopart_seq_record_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 0;
 insert into table alltypesaggmultifilesnopart_seq_record_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 1;
 insert into table alltypesaggmultifilesnopart_seq_record_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 2;
 insert into table alltypesaggmultifilesnopart_seq_record_snap   SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col FROM alltypesagg where id % 4 = 3;
 
+SET hive.exec.compress.output=false; 
+set mapred.output.compression.type=NONE;
+SET mapred.output.compression.codec=NONE;
 -- Create multiple files for alltypesaggmultifiles (hdfs/rc/text)
 INSERT INTO TABLE alltypesaggmultifiles partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
 INSERT INTO TABLE alltypesaggmultifiles_rc partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_rc partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_rc partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_rc partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
 INSERT INTO TABLE alltypesaggmultifiles_text partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_text partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_text partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
@@ -996,34 +1308,73 @@ INSERT INTO TABLE alltypesaggmultifiles_seq partition (year, month, day) SELECT
 INSERT INTO TABLE alltypesaggmultifiles_seq partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET hive.exec.compress.output=true; 
+set mapred.output.compression.type=BLOCK;
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
+INSERT INTO TABLE alltypesaggmultifiles_rc_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
+INSERT INTO TABLE alltypesaggmultifiles_rc_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
+INSERT INTO TABLE alltypesaggmultifiles_rc_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
+INSERT INTO TABLE alltypesaggmultifiles_rc_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
 INSERT INTO TABLE alltypesaggmultifiles_seq_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
+INSERT INTO TABLE alltypesaggmultifiles_rc_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
+INSERT INTO TABLE alltypesaggmultifiles_rc_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
+INSERT INTO TABLE alltypesaggmultifiles_rc_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
+INSERT INTO TABLE alltypesaggmultifiles_rc_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
 INSERT INTO TABLE alltypesaggmultifiles_seq_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
+INSERT INTO TABLE alltypesaggmultifiles_rc_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
+INSERT INTO TABLE alltypesaggmultifiles_rc_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
+INSERT INTO TABLE alltypesaggmultifiles_rc_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
+INSERT INTO TABLE alltypesaggmultifiles_rc_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
 INSERT INTO TABLE alltypesaggmultifiles_seq_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
+INSERT INTO TABLE alltypesaggmultifiles_rc_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
+INSERT INTO TABLE alltypesaggmultifiles_rc_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
+INSERT INTO TABLE alltypesaggmultifiles_rc_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
+INSERT INTO TABLE alltypesaggmultifiles_rc_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
 INSERT INTO TABLE alltypesaggmultifiles_seq_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+set mapred.output.compression.type=RECORD;
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_def partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_gzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_bzip partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 3;
+
+SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 0;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 1;
 INSERT INTO TABLE alltypesaggmultifiles_seq_record_snap partition (year, month, day) SELECT id, bool_col, tinyint_col, smallint_col, int_col, bigint_col, float_col, double_col, date_string_col, string_col, timestamp_col, year, month, day FROM alltypesagg where id % 4 = 2;

testdata/bin/load.sql

@@ -117,6 +117,34 @@ LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypes
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserror_rc PARTITION(year=2009, month=2);
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserror_rc PARTITION(year=2009, month=3);
 
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_def/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_def PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_def/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_def PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_def/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_def PARTITION(year=2009, month=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_def/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_def PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_def/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_def PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_def/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_def PARTITION(year=2009, month=3);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_gzip/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_gzip PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_gzip/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_gzip PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_gzip/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_gzip PARTITION(year=2009, month=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_gzip/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_gzip PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_gzip/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_gzip PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_gzip/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_gzip PARTITION(year=2009, month=3);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_bzip/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_bzip PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_bzip/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_bzip PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_bzip/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_bzip PARTITION(year=2009, month=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_bzip/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_bzip PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_bzip/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_bzip PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_bzip/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_bzip PARTITION(year=2009, month=3);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_snap/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_snap PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_snap/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_snap PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserrornonulls_rc_snap/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserrornonulls_rc_snap PARTITION(year=2009, month=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_snap/year=2009/month=1/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_snap PARTITION(year=2009, month=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_snap/year=2009/month=2/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_snap PARTITION(year=2009, month=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypeserror_rc_snap/year=2009/month=3/000000_0' OVERWRITE INTO TABLE alltypeserror_rc_snap PARTITION(year=2009, month=3);
+
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc PARTITION(year=2010, month=1, day=1);
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc PARTITION(year=2010, month=1, day=10);
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc PARTITION(year=2010, month=1, day=2);
@@ -139,6 +167,94 @@ LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypes
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesagg_rc PARTITION(year=2010, month=1, day=8);
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesagg_rc PARTITION(year=2010, month=1, day=9);
 
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_def/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_def PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_def/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_def PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_bzip/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_bzip PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_bzip/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_bzip PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_snap/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_snap PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_snap/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_snap PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_rc_gzip/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_rc_gzip PARTITION(year=2010, month=1, day=9);
+
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=1);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=10);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=2/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=2);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=3/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=3);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=4/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=4);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=5/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=5);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=6/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=6);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=7/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=7);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=8/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=8);
+LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesagg_rc_gzip/year=2010/month=1/day=9/000000_0' OVERWRITE INTO TABLE alltypesagg_rc_gzip PARTITION(year=2010, month=1, day=9);
+
 -- Load all the sequence databases form saved files
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_seq_bzip/year=2010/month=1/day=10/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_seq_bzip PARTITION(year=2010, month=1, day=10);
 LOAD DATA LOCAL INPATH '${env:IMPALA_HOME}/testdata/data/test-warehouse/alltypesaggnonulls_seq_bzip/year=2010/month=1/day=1/000000_0' OVERWRITE INTO TABLE alltypesaggnonulls_seq_bzip PARTITION(year=2010, month=1, day=1);